AI Evaluations for Everyone: Bringing Human-Centered AI Evaluation to LinkedIn Learning

Co-authors: Mala Kumar, Theodora Skeadas, Annie Brown, Maria di Fonzo

Watch the course now! Available for free until September 30, 2026.

On June 30, the Humane Intelligence LinkedIn Learning course titled, “AI Evaluations for Everyone: How Non-Engineers Can Build Better AI Systems”, will go live! We created this course around the simple idea that AI evaluations should be accessible to everyone, regardless of technical skill. 

Learners will get an overview of AI model- and system-level evaluations, and why they are important for both AI safety and performance. We’ll then take learners through how we map a problem space and how we set up an AI red teaming evaluation. By the end of the course, through a demo, a sample policy, and guidance around AI red teaming design decisions, exploits, and vulnerability types, learners will have their own sample AI red teaming project that they can present to their colleagues, or showcase on their website, portfolio, and to prospective employers.

Why this Course?

As generative AI becomes embedded across workplaces, governments, schools, nonprofits, and public services, organizations are increasingly grappling with questions about whether AI systems are safe, effective, and fit-for-purpose. Addressing those questions require perspectives beyond machine learning and software engineering. Strong AI evaluations require human subject-matter expertise and lived experience, as well as a range of professional skills – from policy professionals to project managers, to researchers and more. This course is designed to close the gap for those who are critical to AI model and system deployment decisions, but don’t necessarily have the technical skills required of other AI evaluation training options. It’s also a great way for more technically (in terms of technology) inclined professionals to understand how non-technical disciplines fit into AI evaluations. 

Course Overview

AI Evaluations for Everyone is a 25-chapter course that introduces learners to the foundations of AI evaluation through Humane Intelligence’s human-centered methodology. Mala, our Executive Director, is the first of three course instructors. She’ll start by helping learners understand the current landscape of AI evaluations by drawing parallels to other disciplines and using a few analogies. This will give the learner an overview of how we approach our work at Humane Intelligence, and draws on our years of experience leading AI red teaming evaluations, bias bounties, and contextual evaluations. And yes, do briefly mention benchmarks.

Next, Mala will cover the idea of defining a problem space, meaning what is in- or out-of-scope of an AI evaluation. As we wrote about in March, Humane Intelligence is developing a new ontology / knowledge graph methodology for contextual AI evaluations and AI red teaming. Related, Mala will take learners of our LinkedIn Learning course through the basics of using a taxonomy versus an ontology to define an AI evaluation. Before handing the course off to our second instructor, Mala will help learners set up their project that they will complete throughout the course.

“One of the most under appreciated aspects about a generative AI-enabled future is that people from every walk of life are needed to get this right. I’m excited that our LinkedIn Learning course will clearly lay out one way that human expertise and lived experience can be incorporated in cutting edge technology.”

~ Mala Kumar, Executive Director, Humane Intelligence.

Annie, our Bias Bounty Data Scientist and the CEO of our close partner organization, Reliabl, is our second instructor. She will talk about the spectrum of fully automated to fully human AI evaluations. Annie will also do a deeper dive into taxonomies and ontologies and their role in contextual evaluations. As an annotations expert, Annie will then talk about metadata classifiers and labeling workflows, and why these are so important to building, evaluating, and fine-tuning AI models and systems.

“This course positions AI evaluation as a core component of the AI development pipeline, showing how human expertise and lived experience can make AI more inclusive and more effective.”

~ Annie Brown, Bias Bounty Data Scientist, Humane Intelligence

 

Theo, our Head of Red Teaming, is our third instructor. She will take learners into a detailed understanding of what can be evaluated in AI models and systems, including bias, hallucinations, factuality, misdirection and security. For the chapter on bias, Theo will talk through our Humane Intelligence work with UNESCO. The factuality chapter focuses on our work with the U.S. National Institute for Standards and Technology (NIST). Theo will talk about our work at DEFCON for the misdirection chapter, and finally, the Australian Information Security Association for the chapter on security.

“Our global, multi-sectoral red teaming work serves as the foundation for our instruction on AI harms across a range of failure modes.”

~ Theodora Skeadas, Head of Red Teaming, Humane Intelligence

 

The course wraps up with Annie’s demo of Reliabl, a ML training and annotations software, Theo’s guidance on how to make sense of AI red teaming results, and Mala’s recommendations for how to leverage the project work to strengthen results at an organization and for those on the job market, how to use it in a job search.

Our LinkedIn Learning course gives concrete details, example use cases, and practical skills related to Humane Intelligence’s mission to break down barriers to AI deployment for social good. By the end, we hope to help learners become a new AI evaluator or strengthen their understanding of how AI evaluations work, why they matter, and how they can contribute to safer, fairer and more accountable AI systems.

A Special Thanks and How to View the Course

Expanding access to AI evaluation requires collaboration, and this LinkedIn Learning course was truly a team effort. We would like to extend our sincere thanks to Natalie Pao, Lori Kallestad, Ryan Turpin, Joni Deluccio, Dione Johnson, and Lalita Segal for their partnership, guidance, and support throughout the production process. Their work helped bring this course from concept to reality.

The course will be available on LinkedIn Learning and accessible to all learners from July 1 through September 30.

If your organization is exploring how to evaluate an AI model or system and would like to speak to us or hire us, please send us an email at info@humane-intelligence.org. Read about our programs and services here.